Вы находитесь на странице: 1из 9

See

discussions, stats, and author profiles for this publication at: https://www.researchgate.net/publication/299678749

New considerations on track-structure


interaction in railway bridges
Chapter July 2010
DOI: 10.1201/b10430-310

CITATIONS

READS

21

3 authors, including:
Manuel Cuadrado
University Carlos III de Madrid
15 PUBLICATIONS 13 CITATIONS
SEE PROFILE

All content following this page was uploaded by Manuel Cuadrado on 19 April 2016.
The user has requested enhancement of the downloaded file.

New considerations on track-structure interaction in railway


bridges
M. Cuadrado Sanguino
Fundacin Caminos de Hierro, Madrid, Spain

P. Gonzlez Requejo
Fundacin Caminos de Hierro, Madrid, Spain

D. Losa Baur
Fundacin Caminos de Hierro, Madrid, Spain

ABSTRACT: Due to longitudinal interaction between track and structure, the presence of
railway bridges may generate added stress on continuous rails and relative displacements between track and deck. Thus an analysis of these phenomena is needed from the structure design
phase on for ensuring proper system behavior.
UIC Leaflet 774-3 and Eurocode EN 1991-2 approach interaction computations by means of
numerical models that idealize the behavior of all elements and actions involved to obtain
stresses and displacements.
This paper describes the conclusions of authors main research and project experience. Issues
not addressed in current standards are described, i.e.:
1. combination of actions: comparison of simplified combination methods versus stepby-step computation methods;
2. considerations on deformations due to shrinkage and creep in concrete pre-stressed
decks;
3. special features of interaction with slab track;
4. track-structure interaction under seismic conditions;
5. application of tailored methodologies to studies on special geometries.
1

COMBINATION OF ACTIONS

1.1 Introduction
A linear combination of stresses and displacements computed for each elementary action (temperature, braking/traction, bending, shrinkage and creep) is not valid for track-structure interaction analysis because the rail-deck interaction model is non-linear, due to the non-linearity of
the force-displacement curve at the track-deck interface. International standards (UIC leaflet
774-3R) admits the combination of results obtained separately for each elementary action
(where braking/traction and deck bending are concomitant) as a simplification of a non-linear
step-by-step computation. This paper describes the comparative analysis of results obtained with
both methods.

1.2 Simplified method of combination of actions


In this simplified method, the introduction on one hand of long-term actions (deformations due
to variations of temperature and eventually to shrinkage and creep in concrete decks), and on the
other hand of traffic actions must be made separately, because the introduction of vertical loads
already modifies the features of the elements that represent the interface between rail and deck
(from unloaded to loaded track), and this requires a step-by-step algorithm.

Traffic actions shall be introduced independently on the model and the results obtained
(stress and displacement) shall be linearly combined with those obtained for long term actions.
Finally, for traffic actions, it is stated that the horizontal forces of braking and traction can
only happen as actions associated with vertical loads and to the corresponding deformation due
to deck vertical bending. Therefore it seems logical that traffic actions be introduced in two
steps. The first step shall consider vertical loads, the second step, where the model has already
experienced strain due to vertical bending, shall include the horizontal forces of braking/traction.
1.3 Step-by-step computation method
As a prerequisite of the step by step calculation, it is necessary to define a law for the longitudinal behavior of the fastening-ballast system as a function of the vertical load. A law like the
one shown in Figure 1 has been defined where the maximum longitudinal load is a function of
the vertical load, introducing a fictitious friction factor :
t
N = 80 kN/m

60 kN/m

increasing N

FN
N=0

F0
2 mm

Figure 1. Resistance of track against relative displacements


with respect to bearing, as a function of vertical load.

The value of maximum force is obtained with the following expression:


FN=F0+N

(1)

where FN = longit. load corresp. to N; F0 = F for N=0; = friction coeff.; N = vertical load
Table1. Longitudinal load values
Ballast poorly maintained: F0 = 12 kN/m Ballast properly maintained:
F80 = 60 kN/m
F80 = 60 kN/m
= 0.60
= 0.50

F0 = 20 kN/m

A full analysis is then possible considering the application of applied strains and subsequently the forces of the train moving along the track model in successive steps.
First of all long-term actions shall be analyzed: these actions may have happened before traffic actions, and the worst-case situations must be found, as in the simplified method.
With regard to traffic actions, they may happen before or after the long-term actions, generally the latter case will be the worst-case situation.
The first difference with the simplified method is that the analysis of the effects of traffic
loads is not made separately before the combination, on the contrary, the traffic loads are introduced on the model deformed by long-term actions.
The second difference with the simplified method lies in how vertical loads are introduced.
These shall act step by step, advancing over the model as to simulate the vehicle progress from
embankment to deck and over the deck itself. At each loading step, the stiffness of non-linear
springs shall be modified depending on vertical load as shown in curve of Figure 1.
Finally, as for the simplified method, with the model already deformed by vertical bending,
horizontal braking/traction forces are introduced.

1.4 Comparison of results


A comparative assessment of the two calculation methods previously described has been made
with several of the test cases stated in UIC Leaflet 774-3R (cases C1-3, D1-3, E1-3 and F1-3). A
single-span bridge 60 m long is analyzed, with a restricted bearing in abutment 1 and a sliding
bearing in abutment 2. The actions considered are: increase of deck temperature + 30C; vertical
traffic load of 80 kN/m; braking load of 20 kN/m. The calculation has been carried out for a
train running from abutment 1 to abutment 2, where the train head is located over abutment 2,
and assuming it will brake at that position. Results show that the linear superposition of applied
strains and traffic loads generates conservative results as regards both stresses in rail and reactions in bearing. The corresponding error is usually small (close to 15% on average and never
greater than 25% in the cases analyzed). On the other hand, in the calculation of both absolute
displacements of deck and relative displacements between track and subgrade or deck , the linear superposition is not conservative. As in the previous case the error is not significant, lower
than 3% on average and never greater than 7% in the cases analyzed.
2 CONSIDERATIONS ON DEFORMATIONS DUE TO SHRINKAGE AND CREEP
2.1 Action due to shrinkage and creep
European and Spanish standards state among other actions to be considered the phenomena of
shrinkage and creep in the deck, although without explicit stating how these have to be considered for the calculation. The overall strains of the deck due to shrinkage and creep can reach
significant values (around 10-3, equivalent to a strain due to a temperature decrease of the
deck of 100 C). Although rapidly decreasing with time, they should be taken into account in all
cases, as they generate stresses in the rail and relative displacements between rail and deck. The
strain curve due to the combined effects of shrinkage and creep can be computed according to
the procedures stated in Spanish standards on concrete (EHE) and represented as a reduction of
equivalent temperature. Once the time elapsed between deck concreting and rail soldering has
been established, the values of this curve allow finding the maximum variation of temperature
equivalent to the strain due to shrinkage and creep at infinite time, i.e. Tt,ret,flu which shall always be a negative value.
2.2 Combination with other actions
The linear combination of stresses and strains obtained by separate computation of each elementary action is not correct and may be excessively conservative if including as an elementary action the strain due to shrinkage and creep.
Thus the authors, although recommending to take into account shrinkage and creep for interaction, consider that the assessment of the worst-case situations for the rail and the most precise
computation possible of the stresses in rail in such situations shall assume concomitance of the
temperature variations arising from strains due to shrinkage and creep and not their linear superposition. The results of this computation shall then linearly superposed with those coming
from the actions of braking/traction and bending (if applying simplified method) or may be used
as initial states of load introduction for the step-by-step method.
Consequently, without any expansion device and considering only the variations of temperature of the deck , the worst-case assumptions will be as follows:
H1) Tt (Tt,ret,flu = 0)
H2) -Tt + (Tt,ret-flu < 0)
On the other hand, considering expansion devices, the variations of temperature of the rail
and deck are both taken into account and thus the worst-case situations will be more than two as
previously described. Therefore it shall be necessary to find several combinations of variations
of deck temperature, eventually including the equivalent variations of temperature from strain
due to shrinkage and creep, and variations of temperature of rail in order to fully assess all
worst-case assumptions.

3 SPECIAL FEATURES OF INTERACTION WITH SLAB TRACK

3.1 Introduction
The parameters of computation models and the verifications to be made for the analysis of
track-structure interaction are clearly defined for ballasted tracks. However, this is not the case
for slab track. On the other hand the different behaviour of both systems leads to design considerations that can be quite different for both types of superstructure. In this paper will be presented the definition of specific points for slab track systems analysis that are not sufficiently
detailed in standards, the relevant models and computations made for representative situations
and the comparison of results in both types with emphasis on design considerations.
3.2 Limitation of stresses
The normative limits of rail stress are established for ballasted track and shall be redefined for
slab track.
For ballasted track, requirements include the risk of track buckling for the usual neutral temperatures. However as this risk does not exist in slab track, the distinction between limit compressive and tensile stress existing for ballasted tracks must disappear for slab track. This is actually the criterion adopted by German standards and by the UIC, where stress limitations on
rail are as follows:
Table 2: Stress limitations on rail
Ballasted track
tensile
s92 MPa
compressive
s72 MPa

Slab track
s92 MPa
s92 MPa

Three types of viaduct have been analysed:


1. Continuous short viaduct, length 100 m, fixed bearing at one abutment (parametrized stiffness) and sliding bearings at the other.
2. Continuous long viaduct, length 800 m, fixed bearing at one abutment (parametrized stiffness) and sliding bearings at the other.
3. statically-determinate 4-span viaduct with parametrized span length, fixed-sliding bearings at
each span (parametrized stiffness).
The mechanical parameters of deck cross-section to be considered for assessing behaviour
against longitudinal interaction are: area, bending inertia, thickness and heighth of center of
gravity. The values used for the 3 viaducts have been derived from an analysis of actual viaducts belonging to HS lines (singular viaducts have not been analysed).
The following can be deduced from the interaction analysis of the different cases:
For short continuous viaducts, the installation of slab track with normal fastenings on continuous rail is a design restriction, as it leads to a reduction of the maximum expansion
length (decrease from 100 to 115 m for ballasted track down to 65 and 70 m for slab track).
However this restriction can be circumvented by installing special sliding fastenings with a
maximum longitudinal resistance limited to 23 kN/(meter of track).
For long continuous viaducts, where an expansion device must be installed at the deck mobile
end, slab track should not mean a design restriction, as the maximum deformation lengths
(depending on the maximum absolute displacement of deck against braking/traction action at
the end with expansion device) are independent of the type of track.
For isostatic viaducts, although there is also a reduction of the maximum dilatable length for
slab track with normal fastenings, these lengths of about 50 m can be considered excessive
for this type of viaducts and therefore this decrease should not mean a design restriction.
3.3 Limitations of displacements
Criteria for assessing limit values for slab track shall depend on tensile forces at the fastenings.

The German standard for slab track states that the maximum tensile force on fastenings on
both sides of the expansion device shall not be greater than the force that decompresses the railpad, which depends on the fastening and can be determined with normalized tests on clamping
force. On the other hand normative limitations on deck rotation at expansion devices are valid
for ballasted track or for ballasted track at one side of the expansion device and slab track at the
other side, while a specific assessment must be made for slab track on both sides of the expansion device.
UIC Leaflet 776-3 includes design criteria on limitations on rail stress, stating that when there
are direct fastenings at both sides of the expansion device, the angular variation and uplift
should be limited in order not to exceed a stress of 80 N/mm2 in the rail, considering the position and actual stiffness of the fastenings. For a UIC60-type rail, this maximum stress is equivalent to a maximum bending moment of 26,8 kNm.
3.3.1 Computation of bending moments on rail and tensile force on fastenings
A 2D model has been implemented with an off-the-shelf FEM software package (See Figure 2).
Rail
Rigid rail-sleeper union
Sleeper

Fastening
Slab track

Deck end-bearing separation


Spring-compression only

Figure 2. Schematic of the model

Tensile behaviour of fastenings has been deduced from normalized tests of clamping force.
3.3.2 Conclusions
As a general conclusion the limitations on rotation and maximum dip stated in standards for the
case of slab track on one side of the expansion device and ballasted track on the other are consistent with the criteria of maximum bending moments on rail and maximum tensile forces on
fastenings, and they are also applicable to the case where there is slab track on both sides of the
expansion device.
Compliance with these criteria will depend on vertical deformability of fastenings and in
some cases a specific study should be needed for assessing limit values, maybe more restrictive.
4 TRACK-STRUCTURE INTERACTION UNDER SEISMIC CONDITIONS
In high-seismicity zones, almost all bridge components are designed with seismic loads in mind.
Furthermore if compared with viaducts in non-seismic zones, design considerations have an impact not only on the volume of material but may radically influence structural design.
Generally speaking, track-structure interaction design considerations are stricter for seismic
conditions. In order to ensure proper viaduct structural behaviour against braking/traction horizontal actions, limiting deck displacements and stresses on rail, fixed points must be established
with sufficient stiffness. Furthermore, as railway bridges tend to be heavier due to requirements
of in-service deformability, seismic forces increase. The requirements on stiffness of fixed
points of deck mean higher structure vibration natural frequencies and thus greater seismic factors.

Therefore seismic considerations may be most important, especially for long viaducts and
even restrict the type of viaduct.
Against a large seismic action viaduct design should be based on a criterion of seismic stress
balance over piers and abutments compatible with requirements of track-structure interaction.
Longitudinal action is the most difficult analysis, especially for long viaducts, due to the coupling of three actions of different nature: the seismic action itself, strains imposed by temperature variations and rheologic effects and braking/traction action. Thus track-structure interaction
is a key requisite for selecting a structural type, especially as regards longitudinal continuity of
deck, location of fixed points to counteract temperature effects and fixed points for resistance to
stress and limitation of displacements due to braking and traction. Main variables will thus be
bridge dilatable length(s), and deformable deck length, depending on longitudinal continuity of
deck and fixed point location, where the difference between dilatable and deformable length is
that fixed points may not be the same for slow deformations (temperature, rheologic effects) or
for fast ones (braking/traction, seismic action) because of installed STUs or dampers.
There are two main options:
continuous deck with fixed points that define dilatable and deformable length of deck,
where expansion devices may be eventually installed.
isostatic spans without longitudinal continuity, with fixed points on each span, where
interaction issues may be addressed without needing expansion devices.
However, for high seismicity zones, deck continuity issues should take into account seismic
actions and their coupling with braking/traction and imposed strains, and also relative displacements of deck.
4.1 Continuous deck
A continuous deck implies a large mass fully excited by seismic action (deck plus total viaduct
dead weight). Longitudinally, there are four types of resistance schemes to the seismic force associated with the vibration of this mass:
a) In the first solution, usually for low-seismicity areas or small deck lengths, a single fixed
point is located at one of the abutments and implying, depending on deck length, the installation
of expansion devices at the other abutment. The main drawback is that in high-seismicity areas
the fixed abutment shall support the whole seismic action.
b) The second solution uses the ductility of piers. In this case, deck shall be free from longitudinal restraint at abutments, thus allowing displacements of the head of piers needed for plastification of base section. Usually expansion devices are installed at both abutments.
This solution also implies transmission of the whole longitudinal action (seismic plus braking/traction) to piers, without help from abutments, and the action shall be spread over a maximum number of piers for better distribution of forces, thus two variants can be implemented.
b.1) direct transmission of longitudinal forces to all or a few piers, implying restriction on
deck movement along most of deck and thus large stresses appearing due to imposed strains
(temperature variations and shrinkage/creep).
b.2) transmission of longitudinal forces to all or a few piers, installing STUs (Shock Transmitter Units) on all or a few piers supporting longitudinal forces .
c) transmission of longitudinal forces to abutments installing STUs only at these points,
bearing in mind that these devices dont dissipate energy and just block the seismic action
which will be fully transmitted to abutments. STU performance against braking force is such
that part of this action is transmitted to piers.
d) installing dampers for absorbing part of the seismic action.
4.2 Isostatic deck
Isostatic spans without longitudinal continuity need fixed points on each span.
Usually in low seismicity areas a 4-bearing system is installed, longitudinally fixing the deck
to pier with restricted bearings while the other end is free with sliding bearings, as shown on the
left in Figure 3.

Fixed bearing

Special fixed bearing


Guided sliding bearing (slow mov.)

Guided sliding bearing

Free sliding bearing


Free sliding bearing

Figure 3: Bearing system for isostatic span bridge

For high seismicity areas the abovementioned 4-bearing system raises some issues.
First of all, relative displacement between contiguous decks is only restricted by the stiffness
of pier and continuous rail. Due to track-structure interaction limitations on this relative displacement, pier head stiffness has to be sufficiently large. Furthermore, in high seismicity areas,
although not required by standards, track stability may be ensured for a moderately large earthquake, corresponding to a lower return period than for design earthquake. In this case it may be
difficult to get proper pier stiffness for limiting relative deck displacements. Transverse relative
displacements between successive piers may also produce deck rotations about the vertical axis.
4-bearing conventional systems imply clamping between deck and pier, therefore these rotations
will in turn induce vertical-axis moments on piers and longitudinal forces on bearings, with
probable impact on viaduct design and cost.
An improvement on bearing system (shown on the right in Figure 3) may allow vertical axis
deck rotations, thus avoiding added stress due to transverse seismic action. If these bearings also
transmit horizontal actions (seismic plus braking/traction) to piers at both ends of deck, relative
displacements between contiguous decks are controlled. If one of these bearings allows slow
displacements (due to temperature variations or shrinkage/creep), dilatable length remains unchanged and continuous rail can be kept without expansion devices without excess stress on rail.
5 APPLICATION OF TAILORED METHODOLOGIES FOR SPECIAL GEOMETRIES

5.1 Introduction
The interaction analysis made for a dual-track viaduct on the Palencia Len line is described, where the structure is a pergola bridge of 94 dual pre-stressed beams T BN-150, with
a length of 18,3 m between bearings and concrete on-site compression slab, with thickness ranging from 0.25 m to 0.39 m. Beam bearings on abutments are elastomeric devices.
5.2 Computation model
A FEM model has been implemented for interaction analysis. Usually for the analysis of conventional viaducts 2D models are used, as shown on Figure 4.
Rail at deck upper face
Non-linear springs between rail and perm. way or deck upper face
Perm. way nodes

Rigid beams between


deck upper face and
axis
Linear spring (stiffness of
abutment&bearings)

Nodes at deck upper face

Rigid beams between


deck axis and bearings
Nodes at
bearing level
Linear spring (stiffness of
pier&bearings)

Figure 4. Schematics of computation model

Axis of deck

This type of models allows computing stresses and displacements on conventional viaducts
but it is not suited for this particular structure, as the longitudinal and transverse deck deformations influence both rail stress and relative displacements between rail and deck. On the other
hand, due to this geometry, there can be situations where one track is on the deck and the other
is still on the embankment and thus they will behave differently.
Then the model has to include geometry on plan of the structure and the actual position of
tracks. On the other hand the interaction analysis takes into account the vertical loads in order to
assess the influence of deck bending on rail stresses. However in this case, the main direction of
bending is the same as the beams and is almost at right angle with track axis and therefore this
effect can be neglected. Thus the computation model can be greatly simplified as a plane model
can be sufficient. On one hand strains and displacements can be considered on the structure
plane and furthermore all elements can be considered on the same plane without taking into
acount the different heights of bearings, axis of deck, deck upper deck and track, as is usually
done to take into account effects of bending.
5.3 Conclusions
A model that includes the geometry on plan of the structure and the actual position of tracks
over it has allowed a justification for not installing expansion device as normative limitations on
rail stress and displacements are met.
6 CONCLUSIONS
This paper summarizes the authors experience on interaction analysis, from the stage of recommendations for railway viaducts to mandatory design requirement reflected in standards. Emphasis is made on a rigorous analysis of these phenomena.
However the standards dont answer to all aspects of analysis methodology: the complexity
of models or computations can be variable (combination of actions, step-by-step computation),
although in all cases the actions considered shall be assessed with a rational approach (e.g.,
shrinkage combined with deck temperature variations), also considering the type of track (ballasted or slab track). Interaction analysis can get more difficult in specific unconventional cases
ie.e in seismic areas, for example if considering damage limitations even for moderate earthquakes, or with special geometries like skew bridges.
The authors have tried to summarize useful recommendations although they consider that the
interaction analysis must depend on a correct assessment of the phenomenon and its importance,
and that it should not be just a complementary part of railway viaduct design.
REFERENCES

Cuadrado, 2004. Consideracin de las deformaciones por retraccin y fluencia en el estudio del fenmeno de interaccin va-tablero en el proyecto de puentes ferroviarios. Revista de Obras Pblicas. Volume N 3.446:45-51.
Cuadrado, 2009. Interaccin Va-Estructura en puentes ferroviarios. Algoritmos de clculo paso a paso.
Revista de Obras Pblicas. Volume N 3499: 39-48.
Eurocode EN 1990. 1990. Annex A2: Basis of structural design. Annex A2: Application for bridges.
Eurocode EN 1991-2. 2003. Acciones en estructuras. Parte 2: Cargas de trfico en puentes.
UIC 2008. Recommendations for Design & Calculation of ballastless track.
UIC 774-3 R. 2000. Interaccin longitudinal va-estructuras. Recomendaciones para los clculos.
UIC 776-3 R. 1989. Dformations des ponts.

View publication stats

Вам также может понравиться